Fig. 3

ROC curve for the diagnosis of DVT in ARDS patients caused by bacterial pneumonia with the combination of SIRS score, serum creatinine level and invasive mechanical ventilation. Multimodal features including SIRS score, serum creatinine level and invasive mechanical ventilation were incorporated into a single model (SSI model) for predicting DVT in patients with ARDS caused by bacterial pneumonia based on the multivariate logistic regression model. The final parameters in the equation are as follows: logit (DVT) =  − 1.620 + 0.255 × SIRS score − 0.013 × serum creatinine (μmol/L) − 1.856 × invasive mechanical ventilation. We generated the validated ROC curve based on the SSI model. Taking the prediction probability of 0.413 as the cut-off point, the model shows satisfactory forecasting ability for predicting DVT (AUC = 0.836; 95% CI: 0.755–0.918; sensitivity: 80.0%; specificity: 74.0%). ARDS, acute respiratory distress syndrome; DVT, deep vein thrombosis; ROC, receiver operating characteristic; AUC, area under the curve; CI, confidence interval